In antononcube/ERTMon-R: Event Records Transformations Monad

library(dplyr)
library(reshape2)
library(ERTMon)

saveDirectoryName <- "../data/WineQualityData/"

Introduction

Get data

dfWineQuality <- read.csv("https://raw.githubusercontent.com/antononcube/MathematicaVsR/master/Data/MathematicaVsR-Data-WineQuality.csv", stringsAsFactors = F)
dim(dfWineQuality)

summary(dfWineQuality)

Convert to medical records

Change "wineQuality" into "Label"

dfWineQuality <-
  dfWineQuality %>%
  mutate( wineQuality = ifelse( wineQuality >= 7, "high", "low") ) %>% 
  mutate( Label = wineQuality ) %>% 
  select(-one_of("wineQuality"))

summary(as.data.frame(unclass(dfWineQuality), stringsAsFactors = T))

Long form

dfWineQualityLongForm <-
  melt( dfWineQuality %>% select(-one_of("Label")) , id.vars = "id" )
names(dfWineQualityLongForm) <- c("EntityID", "Variable", "Value")
dfWineQualityLongForm

dfWineQualityLongForm <-
  dfWineQualityLongForm %>% 
  mutate( ObservationTimeString = "2017-12-04 23:18:48", 
          ObservationTime = 1512429528)

pids <- unique(dfWineQualityLongForm$EntityID)
dfWineEntities <-
  data.frame( EntityID = dfWineQuality$id, 
              Age = floor(runif(n = nrow(dfWineQuality), min = 0, max = 100)), 
              Sex = "U", 
              Label = dfWineQuality$Label, 
              stringsAsFactors = F ) 
dfWineEntities <-
  dfWineEntities %>%
  dplyr::mutate( Label = ifelse( Label == "high", "high", "Non.high") )
dfWineEntities <- melt( dfWineEntities, id.vars = "EntityID")
dfWineEntities <- setNames(dfWineEntities, c("EntityID", "Attribute", "Value"))
dfWineEntities <- dfWineEntities %>% arrange(EntityID)
dfWineEntities

Specifications

#origSpec <- read.csv( "../data/computationSpecification.csv", stringsAsFactors = F)
origSpec <- ERTMonEmptyComputationSpecification( nrow = 1L )
origSpec

outVars <- unique(dfWineQualityLongForm$Variable)
outSpecDF1 <- data.frame( outVars, outVars, "numerical", "NULL", 57600, "Mean", 57600, "Variable", "Mean", "NULL", "NULL", stringsAsFactors = F)
outSpecDF2 <- data.frame( outVars, outVars, "numerical", "NULL", 57600, "OutCnt", 57600, "Age", "Mean", "NULL", "NULL", stringsAsFactors = F)
outSpecDF3 <- data.frame( outVars, outVars, "numerical", "NULL", 57600, "OutFrc", 57600, "Entity", "None", "NULL", "NULL", stringsAsFactors = F)
outSpecDF1 <- setNames( outSpecDF1, names(origSpec) )
outSpecDF2 <- setNames( outSpecDF2, names(origSpec) )
outSpecDF3 <- setNames( outSpecDF3, names(origSpec) )
outSpecDF <- rbind( outSpecDF1, outSpecDF2, outSpecDF3 )
outSpecDF <- rbind( outSpecDF, origSpec[ origSpec$Variable %in% c("Unit", "Label", "Age"), ] )
outSpecDF[ outSpecDF$Variable == "Label", "Critical.label" ] <- "high"
outSpecDF <- outSpecDF[ order(outSpecDF$Variable), ]
outSpecDF

Export

if(params$exportQ) {
  write.csv( x = dfWineEntities, file = paste0( saveDirectoryName, "entityAttributes.csv") )
  write.csv( x = dfWineQualityLongForm, file = paste0( saveDirectoryName, "eventRecords.csv") )
  write.csv( x = outSpecDF, file= paste0( saveDirectoryName, "computationSpecification.csv" ), row.names = F)
}